searchkick 4.4.0 → 5.3.1
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/CHANGELOG.md +160 -3
- data/LICENSE.txt +1 -1
- data/README.md +567 -421
- data/lib/searchkick/bulk_reindex_job.rb +12 -8
- data/lib/searchkick/controller_runtime.rb +40 -0
- data/lib/searchkick/index.rb +167 -74
- data/lib/searchkick/index_cache.rb +30 -0
- data/lib/searchkick/index_options.rb +465 -404
- data/lib/searchkick/indexer.rb +15 -8
- data/lib/searchkick/log_subscriber.rb +57 -0
- data/lib/searchkick/middleware.rb +9 -2
- data/lib/searchkick/model.rb +50 -51
- data/lib/searchkick/process_batch_job.rb +9 -25
- data/lib/searchkick/process_queue_job.rb +4 -3
- data/lib/searchkick/query.rb +106 -77
- data/lib/searchkick/record_data.rb +1 -1
- data/lib/searchkick/record_indexer.rb +136 -51
- data/lib/searchkick/reindex_queue.rb +51 -9
- data/lib/searchkick/reindex_v2_job.rb +10 -34
- data/lib/searchkick/relation.rb +247 -0
- data/lib/searchkick/relation_indexer.rb +155 -0
- data/lib/searchkick/results.rb +131 -96
- data/lib/searchkick/version.rb +1 -1
- data/lib/searchkick/where.rb +11 -0
- data/lib/searchkick.rb +202 -96
- data/lib/tasks/searchkick.rake +14 -10
- metadata +18 -85
- data/CONTRIBUTING.md +0 -53
- data/lib/searchkick/bulk_indexer.rb +0 -173
- data/lib/searchkick/logging.rb +0 -246
@@ -5,28 +5,53 @@ module Searchkick
|
|
5
5
|
def initialize(name)
|
6
6
|
@name = name
|
7
7
|
|
8
|
-
raise
|
8
|
+
raise Error, "Searchkick.redis not set" unless Searchkick.redis
|
9
9
|
end
|
10
10
|
|
11
|
-
|
12
|
-
|
11
|
+
# supports single and multiple ids
|
12
|
+
def push(record_ids)
|
13
|
+
Searchkick.with_redis { |r| r.call("LPUSH", redis_key, record_ids) }
|
14
|
+
end
|
15
|
+
|
16
|
+
def push_records(records)
|
17
|
+
record_ids =
|
18
|
+
records.map do |record|
|
19
|
+
# always pass routing in case record is deleted
|
20
|
+
# before the queue job runs
|
21
|
+
if record.respond_to?(:search_routing)
|
22
|
+
routing = record.search_routing
|
23
|
+
end
|
24
|
+
|
25
|
+
# escape pipe with double pipe
|
26
|
+
value = escape(record.id.to_s)
|
27
|
+
value = "#{value}|#{escape(routing)}" if routing
|
28
|
+
value
|
29
|
+
end
|
30
|
+
|
31
|
+
push(record_ids)
|
13
32
|
end
|
14
33
|
|
15
34
|
# TODO use reliable queuing
|
16
35
|
def reserve(limit: 1000)
|
17
|
-
|
18
|
-
|
19
|
-
|
36
|
+
if supports_rpop_with_count?
|
37
|
+
Searchkick.with_redis { |r| r.call("RPOP", redis_key, limit) }.to_a
|
38
|
+
else
|
39
|
+
record_ids = []
|
40
|
+
Searchkick.with_redis do |r|
|
41
|
+
while record_ids.size < limit && (record_id = r.call("RPOP", redis_key))
|
42
|
+
record_ids << record_id
|
43
|
+
end
|
44
|
+
end
|
45
|
+
record_ids
|
20
46
|
end
|
21
|
-
record_ids.to_a
|
22
47
|
end
|
23
48
|
|
24
49
|
def clear
|
25
|
-
Searchkick.with_redis { |r| r.
|
50
|
+
Searchkick.with_redis { |r| r.call("DEL", redis_key) }
|
26
51
|
end
|
27
52
|
|
28
53
|
def length
|
29
|
-
Searchkick.with_redis { |r| r.
|
54
|
+
Searchkick.with_redis { |r| r.call("LLEN", redis_key) }
|
30
55
|
end
|
31
56
|
|
32
57
|
private
|
@@ -34,5 +59,22 @@ module Searchkick
|
|
34
59
|
def redis_key
|
35
60
|
"searchkick:reindex_queue:#{name}"
|
36
61
|
end
|
62
|
+
|
63
|
+
def supports_rpop_with_count?
|
64
|
+
redis_version >= Gem::Version.new("6.2")
|
65
|
+
end
|
66
|
+
|
67
|
+
def redis_version
|
68
|
+
@redis_version ||=
|
69
|
+
Searchkick.with_redis do |r|
|
70
|
+
info = r.call("INFO")
|
71
|
+
matches = /redis_version:(\S+)/.match(info)
|
72
|
+
Gem::Version.new(matches[1])
|
73
|
+
end
|
74
|
+
end
|
75
|
+
|
76
|
+
def escape(value)
|
77
|
+
value.to_s.gsub("|", "||")
|
78
|
+
end
|
37
79
|
end
|
38
80
|
end
|
@@ -1,41 +1,17 @@
|
|
1
1
|
module Searchkick
|
2
2
|
class ReindexV2Job < ActiveJob::Base
|
3
|
-
RECORD_NOT_FOUND_CLASSES = [
|
4
|
-
"ActiveRecord::RecordNotFound",
|
5
|
-
"Mongoid::Errors::DocumentNotFound",
|
6
|
-
"NoBrainer::Error::DocumentNotFound",
|
7
|
-
"Cequel::Record::RecordNotFound"
|
8
|
-
]
|
9
|
-
|
10
3
|
queue_as { Searchkick.queue_name }
|
11
4
|
|
12
|
-
def perform(
|
13
|
-
model =
|
14
|
-
|
15
|
-
|
16
|
-
|
17
|
-
|
18
|
-
|
19
|
-
|
20
|
-
|
21
|
-
|
22
|
-
# check by name rather than rescue directly so we don't need
|
23
|
-
# to determine which classes are defined
|
24
|
-
raise e unless RECORD_NOT_FOUND_CLASSES.include?(e.class.name)
|
25
|
-
nil
|
26
|
-
end
|
27
|
-
|
28
|
-
unless record
|
29
|
-
record = model.new
|
30
|
-
record.id = id
|
31
|
-
if routing
|
32
|
-
record.define_singleton_method(:search_routing) do
|
33
|
-
routing
|
34
|
-
end
|
35
|
-
end
|
36
|
-
end
|
37
|
-
|
38
|
-
RecordIndexer.new(record).reindex(method_name, mode: :inline)
|
5
|
+
def perform(class_name, id, method_name = nil, routing: nil, index_name: nil)
|
6
|
+
model = Searchkick.load_model(class_name, allow_child: true)
|
7
|
+
index = model.searchkick_index(name: index_name)
|
8
|
+
# use should_index? to decide whether to index (not default scope)
|
9
|
+
# just like saving inline
|
10
|
+
# could use Searchkick.scope() in future
|
11
|
+
# but keep for now for backwards compatibility
|
12
|
+
model = model.unscoped if model.respond_to?(:unscoped)
|
13
|
+
items = [{id: id, routing: routing}]
|
14
|
+
RecordIndexer.new(index).reindex_items(model, items, method_name: method_name, single: true)
|
39
15
|
end
|
40
16
|
end
|
41
17
|
end
|
@@ -0,0 +1,247 @@
|
|
1
|
+
module Searchkick
|
2
|
+
class Relation
|
3
|
+
NO_DEFAULT_VALUE = Object.new
|
4
|
+
|
5
|
+
# note: modifying body directly is not supported
|
6
|
+
# and has no impact on query after being executed
|
7
|
+
# TODO freeze body object?
|
8
|
+
delegate :body, :params, to: :query
|
9
|
+
delegate_missing_to :private_execute
|
10
|
+
|
11
|
+
attr_reader :model
|
12
|
+
alias_method :klass, :model
|
13
|
+
|
14
|
+
def initialize(model, term = "*", **options)
|
15
|
+
@model = model
|
16
|
+
@term = term
|
17
|
+
@options = options
|
18
|
+
|
19
|
+
# generate query to validate options
|
20
|
+
query
|
21
|
+
end
|
22
|
+
|
23
|
+
# same as Active Record
|
24
|
+
def inspect
|
25
|
+
entries = results.first(11).map!(&:inspect)
|
26
|
+
entries[10] = "..." if entries.size == 11
|
27
|
+
"#<#{self.class.name} [#{entries.join(', ')}]>"
|
28
|
+
end
|
29
|
+
|
30
|
+
def execute
|
31
|
+
Searchkick.warn("The execute method is no longer needed")
|
32
|
+
load
|
33
|
+
end
|
34
|
+
|
35
|
+
# experimental
|
36
|
+
def limit(value)
|
37
|
+
clone.limit!(value)
|
38
|
+
end
|
39
|
+
|
40
|
+
# experimental
|
41
|
+
def limit!(value)
|
42
|
+
check_loaded
|
43
|
+
@options[:limit] = value
|
44
|
+
self
|
45
|
+
end
|
46
|
+
|
47
|
+
# experimental
|
48
|
+
def offset(value = NO_DEFAULT_VALUE)
|
49
|
+
# TODO remove in Searchkick 6
|
50
|
+
if value == NO_DEFAULT_VALUE
|
51
|
+
private_execute.offset
|
52
|
+
else
|
53
|
+
clone.offset!(value)
|
54
|
+
end
|
55
|
+
end
|
56
|
+
|
57
|
+
# experimental
|
58
|
+
def offset!(value)
|
59
|
+
check_loaded
|
60
|
+
@options[:offset] = value
|
61
|
+
self
|
62
|
+
end
|
63
|
+
|
64
|
+
# experimental
|
65
|
+
def page(value)
|
66
|
+
clone.page!(value)
|
67
|
+
end
|
68
|
+
|
69
|
+
# experimental
|
70
|
+
def page!(value)
|
71
|
+
check_loaded
|
72
|
+
@options[:page] = value
|
73
|
+
self
|
74
|
+
end
|
75
|
+
|
76
|
+
# experimental
|
77
|
+
def per_page(value = NO_DEFAULT_VALUE)
|
78
|
+
# TODO remove in Searchkick 6
|
79
|
+
if value == NO_DEFAULT_VALUE
|
80
|
+
private_execute.per_page
|
81
|
+
else
|
82
|
+
clone.per_page!(value)
|
83
|
+
end
|
84
|
+
end
|
85
|
+
|
86
|
+
# experimental
|
87
|
+
def per_page!(value)
|
88
|
+
check_loaded
|
89
|
+
@options[:per_page] = value
|
90
|
+
self
|
91
|
+
end
|
92
|
+
|
93
|
+
# experimental
|
94
|
+
def where(value = NO_DEFAULT_VALUE)
|
95
|
+
if value == NO_DEFAULT_VALUE
|
96
|
+
Where.new(self)
|
97
|
+
else
|
98
|
+
clone.where!(value)
|
99
|
+
end
|
100
|
+
end
|
101
|
+
|
102
|
+
# experimental
|
103
|
+
def where!(value)
|
104
|
+
check_loaded
|
105
|
+
if @options[:where]
|
106
|
+
@options[:where] = {_and: [@options[:where], ensure_permitted(value)]}
|
107
|
+
else
|
108
|
+
@options[:where] = ensure_permitted(value)
|
109
|
+
end
|
110
|
+
self
|
111
|
+
end
|
112
|
+
|
113
|
+
# experimental
|
114
|
+
def rewhere(value)
|
115
|
+
clone.rewhere!(value)
|
116
|
+
end
|
117
|
+
|
118
|
+
# experimental
|
119
|
+
def rewhere!(value)
|
120
|
+
check_loaded
|
121
|
+
@options[:where] = ensure_permitted(value)
|
122
|
+
self
|
123
|
+
end
|
124
|
+
|
125
|
+
# experimental
|
126
|
+
def order(*values)
|
127
|
+
clone.order!(*values)
|
128
|
+
end
|
129
|
+
|
130
|
+
# experimental
|
131
|
+
def order!(*values)
|
132
|
+
values = values.first if values.size == 1 && values.first.is_a?(Array)
|
133
|
+
check_loaded
|
134
|
+
(@options[:order] ||= []).concat(values)
|
135
|
+
self
|
136
|
+
end
|
137
|
+
|
138
|
+
# experimental
|
139
|
+
def reorder(*values)
|
140
|
+
clone.reorder!(*values)
|
141
|
+
end
|
142
|
+
|
143
|
+
# experimental
|
144
|
+
def reorder!(*values)
|
145
|
+
check_loaded
|
146
|
+
@options[:order] = values
|
147
|
+
self
|
148
|
+
end
|
149
|
+
|
150
|
+
# experimental
|
151
|
+
def select(*values, &block)
|
152
|
+
if block_given?
|
153
|
+
private_execute.select(*values, &block)
|
154
|
+
else
|
155
|
+
clone.select!(*values)
|
156
|
+
end
|
157
|
+
end
|
158
|
+
|
159
|
+
# experimental
|
160
|
+
def select!(*values)
|
161
|
+
check_loaded
|
162
|
+
(@options[:select] ||= []).concat(values)
|
163
|
+
self
|
164
|
+
end
|
165
|
+
|
166
|
+
# experimental
|
167
|
+
def reselect(*values)
|
168
|
+
clone.reselect!(*values)
|
169
|
+
end
|
170
|
+
|
171
|
+
# experimental
|
172
|
+
def reselect!(*values)
|
173
|
+
check_loaded
|
174
|
+
@options[:select] = values
|
175
|
+
self
|
176
|
+
end
|
177
|
+
|
178
|
+
# experimental
|
179
|
+
def includes(*values)
|
180
|
+
clone.includes!(*values)
|
181
|
+
end
|
182
|
+
|
183
|
+
# experimental
|
184
|
+
def includes!(*values)
|
185
|
+
check_loaded
|
186
|
+
(@options[:includes] ||= []).concat(values)
|
187
|
+
self
|
188
|
+
end
|
189
|
+
|
190
|
+
# experimental
|
191
|
+
def only(*keys)
|
192
|
+
Relation.new(@model, @term, **@options.slice(*keys))
|
193
|
+
end
|
194
|
+
|
195
|
+
# experimental
|
196
|
+
def except(*keys)
|
197
|
+
Relation.new(@model, @term, **@options.except(*keys))
|
198
|
+
end
|
199
|
+
|
200
|
+
# experimental
|
201
|
+
def load
|
202
|
+
private_execute
|
203
|
+
self
|
204
|
+
end
|
205
|
+
|
206
|
+
def loaded?
|
207
|
+
!@execute.nil?
|
208
|
+
end
|
209
|
+
|
210
|
+
def respond_to_missing?(method_name, include_all)
|
211
|
+
Results.new(nil, nil, nil).respond_to?(method_name, include_all) || super
|
212
|
+
end
|
213
|
+
|
214
|
+
# TODO uncomment in 6.0
|
215
|
+
# def to_yaml
|
216
|
+
# private_execute.to_a.to_yaml
|
217
|
+
# end
|
218
|
+
|
219
|
+
private
|
220
|
+
|
221
|
+
def private_execute
|
222
|
+
@execute ||= query.execute
|
223
|
+
end
|
224
|
+
|
225
|
+
def query
|
226
|
+
@query ||= Query.new(@model, @term, **@options)
|
227
|
+
end
|
228
|
+
|
229
|
+
def check_loaded
|
230
|
+
raise Error, "Relation loaded" if loaded?
|
231
|
+
|
232
|
+
# reset query since options will change
|
233
|
+
@query = nil
|
234
|
+
end
|
235
|
+
|
236
|
+
# provides *very* basic protection from unfiltered parameters
|
237
|
+
# this is not meant to be comprehensive and may be expanded in the future
|
238
|
+
def ensure_permitted(obj)
|
239
|
+
obj.to_h
|
240
|
+
end
|
241
|
+
|
242
|
+
def initialize_copy(other)
|
243
|
+
super
|
244
|
+
@execute = nil
|
245
|
+
end
|
246
|
+
end
|
247
|
+
end
|
@@ -0,0 +1,155 @@
|
|
1
|
+
module Searchkick
|
2
|
+
class RelationIndexer
|
3
|
+
attr_reader :index
|
4
|
+
|
5
|
+
def initialize(index)
|
6
|
+
@index = index
|
7
|
+
end
|
8
|
+
|
9
|
+
def reindex(relation, mode:, method_name: nil, full: false, resume: false, scope: nil)
|
10
|
+
# apply scopes
|
11
|
+
if scope
|
12
|
+
relation = relation.send(scope)
|
13
|
+
elsif relation.respond_to?(:search_import)
|
14
|
+
relation = relation.search_import
|
15
|
+
end
|
16
|
+
|
17
|
+
# remove unneeded loading for async and queue
|
18
|
+
if mode == :async || mode == :queue
|
19
|
+
if relation.respond_to?(:primary_key)
|
20
|
+
relation = relation.except(:includes, :preload)
|
21
|
+
unless mode == :queue && relation.klass.method_defined?(:search_routing)
|
22
|
+
relation = relation.except(:select).select(relation.primary_key)
|
23
|
+
end
|
24
|
+
elsif relation.respond_to?(:only)
|
25
|
+
unless mode == :queue && relation.klass.method_defined?(:search_routing)
|
26
|
+
relation = relation.only(:_id)
|
27
|
+
end
|
28
|
+
end
|
29
|
+
end
|
30
|
+
|
31
|
+
if mode == :async && full
|
32
|
+
return full_reindex_async(relation)
|
33
|
+
end
|
34
|
+
|
35
|
+
relation = resume_relation(relation) if resume
|
36
|
+
|
37
|
+
reindex_options = {
|
38
|
+
mode: mode,
|
39
|
+
method_name: method_name,
|
40
|
+
full: full
|
41
|
+
}
|
42
|
+
record_indexer = RecordIndexer.new(index)
|
43
|
+
|
44
|
+
in_batches(relation) do |items|
|
45
|
+
record_indexer.reindex(items, **reindex_options)
|
46
|
+
end
|
47
|
+
end
|
48
|
+
|
49
|
+
def batches_left
|
50
|
+
Searchkick.with_redis { |r| r.call("SCARD", batches_key) }
|
51
|
+
end
|
52
|
+
|
53
|
+
def batch_completed(batch_id)
|
54
|
+
Searchkick.with_redis { |r| r.call("SREM", batches_key, [batch_id]) }
|
55
|
+
end
|
56
|
+
|
57
|
+
private
|
58
|
+
|
59
|
+
def resume_relation(relation)
|
60
|
+
if relation.respond_to?(:primary_key)
|
61
|
+
# use total docs instead of max id since there's not a great way
|
62
|
+
# to get the max _id without scripting since it's a string
|
63
|
+
where = relation.arel_table[relation.primary_key].gt(index.total_docs)
|
64
|
+
relation = relation.where(where)
|
65
|
+
else
|
66
|
+
raise Error, "Resume not supported for Mongoid"
|
67
|
+
end
|
68
|
+
end
|
69
|
+
|
70
|
+
def in_batches(relation)
|
71
|
+
if relation.respond_to?(:find_in_batches)
|
72
|
+
klass = relation.klass
|
73
|
+
# remove order to prevent possible warnings
|
74
|
+
relation.except(:order).find_in_batches(batch_size: batch_size) do |batch|
|
75
|
+
# prevent scope from affecting search_data as well as inline jobs
|
76
|
+
# Active Record runs relation calls in scoping block
|
77
|
+
# https://github.com/rails/rails/blob/main/activerecord/lib/active_record/relation/delegation.rb
|
78
|
+
# note: we could probably just call klass.current_scope = nil
|
79
|
+
# anywhere in reindex method (after initial all call),
|
80
|
+
# but this is more cautious
|
81
|
+
previous_scope = klass.current_scope(true)
|
82
|
+
if previous_scope
|
83
|
+
begin
|
84
|
+
klass.current_scope = nil
|
85
|
+
yield batch
|
86
|
+
ensure
|
87
|
+
klass.current_scope = previous_scope
|
88
|
+
end
|
89
|
+
else
|
90
|
+
yield batch
|
91
|
+
end
|
92
|
+
end
|
93
|
+
else
|
94
|
+
klass = relation.klass
|
95
|
+
each_batch(relation, batch_size: batch_size) do |batch|
|
96
|
+
# prevent scope from affecting search_data as well as inline jobs
|
97
|
+
# note: Model.with_scope doesn't always restore scope, so use custom logic
|
98
|
+
previous_scope = Mongoid::Threaded.current_scope(klass)
|
99
|
+
if previous_scope
|
100
|
+
begin
|
101
|
+
Mongoid::Threaded.set_current_scope(nil, klass)
|
102
|
+
yield batch
|
103
|
+
ensure
|
104
|
+
Mongoid::Threaded.set_current_scope(previous_scope, klass)
|
105
|
+
end
|
106
|
+
else
|
107
|
+
yield batch
|
108
|
+
end
|
109
|
+
end
|
110
|
+
end
|
111
|
+
end
|
112
|
+
|
113
|
+
def each_batch(relation, batch_size:)
|
114
|
+
# https://github.com/karmi/tire/blob/master/lib/tire/model/import.rb
|
115
|
+
# use cursor for Mongoid
|
116
|
+
items = []
|
117
|
+
relation.all.each do |item|
|
118
|
+
items << item
|
119
|
+
if items.length == batch_size
|
120
|
+
yield items
|
121
|
+
items = []
|
122
|
+
end
|
123
|
+
end
|
124
|
+
yield items if items.any?
|
125
|
+
end
|
126
|
+
|
127
|
+
def batch_size
|
128
|
+
@batch_size ||= index.options[:batch_size] || 1000
|
129
|
+
end
|
130
|
+
|
131
|
+
def full_reindex_async(relation)
|
132
|
+
batch_id = 1
|
133
|
+
class_name = relation.searchkick_options[:class_name]
|
134
|
+
|
135
|
+
in_batches(relation) do |items|
|
136
|
+
batch_job(class_name, batch_id, items.map(&:id))
|
137
|
+
batch_id += 1
|
138
|
+
end
|
139
|
+
end
|
140
|
+
|
141
|
+
def batch_job(class_name, batch_id, record_ids)
|
142
|
+
Searchkick.with_redis { |r| r.call("SADD", batches_key, [batch_id]) }
|
143
|
+
Searchkick::BulkReindexJob.perform_later(
|
144
|
+
class_name: class_name,
|
145
|
+
index_name: index.name,
|
146
|
+
batch_id: batch_id,
|
147
|
+
record_ids: record_ids.map { |v| v.instance_of?(Integer) ? v : v.to_s }
|
148
|
+
)
|
149
|
+
end
|
150
|
+
|
151
|
+
def batches_key
|
152
|
+
"searchkick:reindex:#{index.name}:batches"
|
153
|
+
end
|
154
|
+
end
|
155
|
+
end
|